ECONOMIC SCENARIOS BASED ON REAL-WORLD PROBABILITIES

A sensible estimation of the value, income and liquidity analysis must be based on economic scenarios which are based on real-world probabilities. Such scenarios may be less structured mathematically and thus cannot be defined in the same precise terms as those used to formulate arbitrage-free models.

Even though there is no common agreement among economists about forecasting tech­niques to define economic scenarios, there are, technically speaking, three distinct methods for constructing economic scenarios:

■ What-if scenarios. These are deterministically based scenarios where an analyst has the freedom to define the future evolution of market risk factors, e.g., a scenario of 1.5% of parallel shift of the yield curve over a one-year horizon can be defined. In practice, analysts may apply scenarios where the prices of risk factors remain constant throughout the simulation horizon. They may use a forward scenario in which all future spot prices are derived by forwarding the currently observed market prices. These scenarios make it possible to compare value and income effects over time.

■ Historical model scenarios. These are based on statistical models which are based on historical data observations. Past stress periods can be a good benchmark for defining such scenarios. Of course, history does not necessarily repeat itself, and then not at the same frequency and magnitude.

■ Stochastic scenarios. These are stochastically driven scenarios based on some economic research of the considered risk factors and algorithmic random scenario generators, e.g., Monte Carlo techniques. The number of generated scenarios is rather large. We consider such forecasted scenarios unbiased, but also difficult to examine at an individual level.

The definition and generation of economic scenarios requires considerable analysis of economic expectations, observations, as well as algorithmic approaches.

6.2